# Bioinspired Auxetic Metastructures Enable Biomechanically Adaptive, Machine Learning-Enhanced Self-Powered Sensing with Ultrahigh Efficiency

## TL;DR

A bioinspired self-powered sensor with machine learning achieves high efficiency and accuracy for wearable applications.

## Contribution

A bioinspired auxetic metastructure improves energy conversion and sensing accuracy in self-powered wearable devices.

## Key findings

- The device achieves 3.2-fold higher bending-mode energy conversion efficiency than non-auxetic controls.
- Integrated machine learning enables 98.7% object recognition accuracy.
- Output voltage reaches 478 V with 13.8% energy conversion efficiency in linear mode.

## Abstract

Bioinspired auxetic triboelectric nanogenerator utilizes negative Poisson’s ratio to resolve interfacial mechanical mismatch. A “conformal self-adaptation” mechanism via synclastic curvature maximizes contact area and signal stability on curvilinear surfaces.The optimized structure achieves a 3.2-fold increase in bending-mode energy conversion efficiency compared to non-auxetic controls, ensuring robust energy harvesting performance under dynamic deformation.An integrated self-powered sensor array coupled with a convolutional neural network deep learning model enables intelligent object recognition with 98.7% accuracy, demonstrating precise human–machine interaction capabilities.

Bioinspired auxetic triboelectric nanogenerator utilizes negative Poisson’s ratio to resolve interfacial mechanical mismatch. A “conformal self-adaptation” mechanism via synclastic curvature maximizes contact area and signal stability on curvilinear surfaces.

The optimized structure achieves a 3.2-fold increase in bending-mode energy conversion efficiency compared to non-auxetic controls, ensuring robust energy harvesting performance under dynamic deformation.

An integrated self-powered sensor array coupled with a convolutional neural network deep learning model enables intelligent object recognition with 98.7% accuracy, demonstrating precise human–machine interaction capabilities.

The online version contains supplementary material available at 10.1007/s40820-026-02125-8.

Self-powered flexible sensors exhibit revolutionary potential in next-generation wearable technologies owing to their exceptional sensitivity and self-sustaining energy harvesting capabilities. Nevertheless, their widespread deployment remains constrained by three fundamental challenges: dynamic mechanical mismatch between biological tissues and rigid devices, suboptimal energy conversion efficiency, and interfacial impedance fluctuation under deformation. Drawing inspiration from the unique negative Poisson’s ratio mesh architecture of lacewing wings, we present a bioinspired auxetic metastructure-engineered triboelectric nanogenerator. This innovative design integrates engineered collagen and micropatterned fluorinated ethylene propylene as triboelectric layers, unified by an auxetic framework with re-entrant hexagonal unit cells interconnected via triangular ligaments. The metastructure enables exceptional lateral expansion under longitudinal strain while simultaneously enhancing structural rigidity and deformation adaptability. This dual functionality effectively minimizes tissue–device mechanical mismatch, thereby significantly improving signal fidelity, sensitivity, and mechanical-to-electrical conversion efficiency during multi-axial deformations. The optimized device achieves remarkable performance metrics, delivering 478 V output voltage with 13.8% energy conversion efficiency in linear configuration, while demonstrating threefold enhanced stability (58 V, 7.58% efficiency) under complex bending compared to conventional designs. Integrated with a convolutional neural network-based machine learning enables exceptional classification accuracy (> 99%) across diverse material recognition tasks, validating its robustness as a next-generation platform for adaptive self-powered wearable sensing.

The online version contains supplementary material available at 10.1007/s40820-026-02125-8.